Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Safety Evaluation on HarmBench (Safety Score %)

4.1ASR

SET

1.36819.80938.2556.691Apr 9, 2026Apr 10, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.04
4.1--
2026.04
5.2--
2026.04
7.2--
2026.04
7.4--
2026.04
7.4--
2026.04
13.6--
2026.04
17--
2026.04
17.9--
2026.04
18.4--
2026.04
20.3--
2026.04
21.8--
2026.04
22.1--
2026.04
23.4--
2026.04
23.5--
2026.04
25.4--
2026.04
25.9--
2026.04
26.9--
2026.04
28.9--
2026.04
29.7--
2026.04
31.8--
2026.04
32.7--
2026.04
33.2--
2026.04
34.6--
2026.04
34.8--
2026.04
36.5-13
2026.04
39.2--
2026.04
41.2--
2026.04
41.2--
2026.04
42-18.5
2026.04
43.7--
2026.04
44.9--
2026.04
46.2--
2026.04
47.3--
2026.04
54.5--
2026.04
55.1--
2026.04
55.1--
2026.04
59-4.5
2026.04
60.8--
2026.04
61.2--
2026.04
67-12.5
2026.04
72.4--
2026.04
72.4--
2026.03
-47.5-
2026.03
-94.2-
2026.03
-57.3-
2026.03
-94.8-
2026.03
-57.1-
2026.03
-66.115.8
2026.03
-71.625.4
2026.03
-69.321.3
2026.03
-74.129.7
2026.03
-75.832.7
2026.03
-87.4-
2026.03
-95.18.8
2026.03
-969.9
2026.03
-95.79.5
2026.03
-96.410.3
2026.03
-96.810.8